Quality-oriented nonlinear dynamic process monitoring method

Abstract

The invention relates to a quality-oriented nonlinear dynamic process monitoring method. The method comprises the following steps of (1) collecting process data and quality data of a nonlinear processunder normal working conditions, constructing modeling data, and standardizing; (2) calculating a standard square sum of typical vectors related to quality and control limits; (3) calculating a standard square sum of principal element vectors not related to quality, squared prediction errors SPEex and control limits; (4) monitoring process variables in the nonlinear dynamic process and quality variable data in real time, calculating standard square sums SPEex statistics, and judging whether or not a failure occurs and whether or not the process failure affects product quality. The method caneffectively filter out process failure alarms which do not affect the product quality, improve the credibility of process monitoring, enhance the logic integrity and practicability of a real-time process monitoring system, and adapt to the requirements of real-time process monitoring of a nonlinear dynamic process; a process failure diagnosis is accurate, the real-time process monitoring efficiency is high, and the adaptation range is wide.

Description

Technical field

[0001] The invention relates to the technical field of chemical production process monitoring, in particular to a quality-oriented nonlinear dynamic process monitoring method. Background technique

[0002] The modern chemical industry's production process is becoming more and more complex, and process safety and reliability are of paramount importance. Process monitoring technology uses measurement data to monitor the running status of the process and alarm for abnormal conditions. Therefore, process monitoring technology is the key to ensuring process safety, stability, and long-term operation. Process variables are often collected online, and quality variables are often obtained offline. The sampling frequency is low and there is a time lag. Considering the real-time nature of process monitoring, traditional data-driven process monitoring methods only use process data to monitor the running status of the process, such as principal component analysis, independe...

Images